Graduation Semester and Year




Document Type


Degree Name

Doctor of Philosophy in Computer Science


Computer Science and Engineering

First Advisor

Mohan Kumar


It is hard to believe that the Internet is now in its adolescent stage. This information age is replete with communication capable, intelligent, sensor equipped devices. Social networks, web services, and global information repositories make a wealth of information available instantly. There exist endless possibilities for creating useable knowledge. Much of what is considered useable knowledge is not directly observable from low level sensory devices. Abstract situations, relationships and activities must be inferred using a variety of techniques that fuse information from multivariate data sources. We refer to this useable knowledge as high level context. Social, physiological, environmental, computational, activity, location and situation are but a few categories of high level context used today. In a general sense, context is any domain specific knowledge relevant to decision making. Low level contexts can be inferred after minimal manipulation and preprocessing of sensor data. High level context is intrinsically more complex. High level context involves many levels of data fusion for inferring high level concepts. The increased dimensionality of representing and reasoning on relationships among contextual components, factoring uncertainty and ignorance, makes it difficult to effectively reason. A research problem in the area of context-aware computing is adaptive and effective high-level context reasoning. Effectiveness refers to the suitability of reasoning methodology for efficiently reasoning and representing the heterogeneous characteristics of context. Adaptive reasoning aides in maintaining context content and quality in the face of dynamic resource availability, degrading reasoning performance and evolving requirements. Context architects are at times challenged; constrained by the limited reasoning provided in the available platforms. Incorporating a generalized hierarchical hybrid reasoning engine, offering variety and optimization for reasoning across heterogeneous complex contexts would provide an effective alternative. Such architecture integrates a variety of configurable reasoning techniques, supporting the modularity of complex high level context. Ultimately, it promotes context reasoning framework reuse, knowledge sharing, and improved context aware application performance. This research proposes novel enabling solutions for adaptive and effective reasoning in pervasive environments. The focus is on middleware solutions for deriving and sustaining high level context, with support for reasoning adaptation and quality maintenance in dynamic pervasive environments. These solutions provided can be used for initiating context inference applications or extending existing architectures for greater reusability. Reuse leads to rapid and innovative context aware application development, a necessary evolution for achieving the vision of ubiquitous computing and beyond.


Computer Sciences | Physical Sciences and Mathematics


Degree granted by The University of Texas at Arlington